Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.user > #265983

Re: new thread, are gigastone ssd's good/bad

Show all headers | View raw

On 1/14/24 13:47, gene heskett wrote:
> Regardless of the sata port/card this 5th SSD is plugged into, it 
> appears to me that rsync is overpowering the SSD's write bandwidth and 
> somewhere between 13.5Gb of data to around 45Gb of data copied, I wind 
> up with a machine that begins to stutter a bit, and finally htop says 
> rsync is down to zero cpu, and maybe one more file locks everything but 
> the mouse up, it still moves but the clock and the rest of the machine 
> has stopped. This may take 5 minutes, maybe 15 but its the same story 5 
> times now. by the time I've rebooted, smartctl give the SSD a clean bill 
> of health. or did til this last try, now it says, for that drive:
> 
> ene@coyote:~/src/klipper-docs$ sudo smartctl -a /dev/sde
> [sudo] password for gene:
> smartctl 7.3 2022-02-28 r5338 [x86_64-linux-6.1.0-17-rt-amd64] (local 
> build)
> ,,,
> SMART Attributes Data Structure revision number: 1
> Vendor Specific SMART Attributes with Thresholds:
> ID# ATTRIBUTE_NAME          FLAG     VALUE WORST THRESH TYPE UPDATED  
> WHEN_FAILED RAW_VALUE
>    1 Raw_Read_Error_Rate     0x0032   100   100   050    Old_age 
> Always       -       0
>    5 Reallocated_Sector_Ct   0x0032   100   100   050    Old_age 
> Always       -       0
>    9 Power_On_Hours          0x0032   100   100   050    Old_age 
> Always       -       912
>   12 Power_Cycle_Count       0x0032   100   100   050    Old_age 
> Always       -       7
> 160 Unknown_Attribute       0x0032   100   100   050    Old_age   Always 
>        -       0
> 161 Unknown_Attribute       0x0033   100   100   050    Pre-fail  Always 
>        -       100
> 163 Unknown_Attribute       0x0032   100   100   050    Old_age   Always 
>        -       10
> 164 Unknown_Attribute       0x0032   100   100   050    Old_age   Always 
>        -       231
> 165 Unknown_Attribute       0x0032   100   100   050    Old_age   Always 
>        -       2
> 166 Unknown_Attribute       0x0032   100   100   050    Old_age   Always 
>        -       0
> 167 Unknown_Attribute       0x0032   100   100   050    Old_age   Always 
>        -       0
> 168 Unknown_Attribute       0x0032   100   100   050    Old_age   Always 
>        -       1500
> 169 Unknown_Attribute       0x0032   100   100   050    Old_age   Always 
>        -       100
> 175 Program_Fail_Count_Chip 0x0032   100   100   050    Old_age   Always 
>        -       0
> 176 Erase_Fail_Count_Chip   0x0032   100   100   050    Old_age   Always 
>        -       0
> 177 Wear_Leveling_Count     0x0032   100   100   050    Old_age   Always 
>        -       0
> 178 Used_Rsvd_Blk_Cnt_Chip  0x0032   100   100   050    Old_age   Always 
>        -       0
> 181 Program_Fail_Cnt_Total  0x0032   100   100   050    Old_age   Always 
>        -       0
> 182 Erase_Fail_Count_Total  0x0032   100   100   050    Old_age   Always 
>        -       0
> 192 Power-Off_Retract_Count 0x0032   100   100   050    Old_age   Always 
>        -       6
> 194 Temperature_Celsius     0x0022   100   100   050    Old_age   Always 
>        -       40
> 195 Hardware_ECC_Recovered  0x0032   100   100   050    Old_age   Always 
>        -       0
> 196 Reallocated_Event_Count 0x0032   100   100   050    Old_age   Always 
>        -       0
> 197 Current_Pending_Sector  0x0032   100   100   050    Old_age   Always 
>        -       0
> 198 Offline_Uncorrectable   0x0032   100   100   050    Old_age   Always 
>        -       0
> 199 UDMA_CRC_Error_Count    0x0032   100   100   050    Old_age   Always 
>        -       2
> 232 Available_Reservd_Space 0x0032   100   100   050    Old_age   Always 
>        -       100
> 241 Total_LBAs_Written      0x0030   100   100   050    Old_age 
> Offline      -       5318
> 242 Total_LBAs_Read         0x0030   100   100   050    Old_age 
> Offline      -       87
> 245 Unknown_Attribute       0x0032   100   100   050    Old_age   Always 
>        -       0


I believe smartctl(8) requests SMART data with binary commands, receives 
binary replies, and must decode those replies based upon details from 
the manufacturer.  I expect that there are time delays between 
smartctl(8) supporting a given drive and when that version of 
smartctl(8) makes it into Debian stable.


Power_On_Hours raw value is 912.  You have stated that this is a 
relatively new drive.  "Old_age" and "Pre-fail' do not make sense.  I 
would ignore them.


Another clue is the multiple "Unknown_Attribute" rows.  This tells me 
that your smartctl(8) does not fully support that Gigastone drive.


The drive is relatively new, so VALUE=100 and WORST=100 for all 
attributes makes sense.



> SMART Error Log Version: 1
> ATA Error Count: 2


Note that "ATA Error Count"=2 matches "UDMA_CRC_Error_Count"=2 above. 
That indicates communications problems between the drive and the 
motherboard or HBA.


Please look for ATA error messages in dmesg and post anything related:

# dmesg | grep -i error


The most likely cause is the SATA cable.  I would install a brand new 
SATA III 6 Gbps cable with locking connectors.


> Error -2 occurred at disk power-on lifetime: 0 hours (0 days + 0 hours)


More clues that smartctl(8) does not fully support that drive.

> SMART Self-test log structure revision number 1
> Num  Test_Description    Status                  Remaining 
> LifeTime(hours)  LBA_of_first_error
> # 1  Extended offline    Completed without error       00%       901     -


The SMART long self-test completed without error, so the drive inards 
are good.


> Is there a way to throttle rsync to the write bandwidth of these SSD's?

Yes:

2024-01-15 00:17:44 dpchrist@laalaa ~/paul-christensen/g6950v
$ man rsync | grep bwlimit
        --bwlimit=RATE           limit socket I/O bandwidth
        --bwlimit=RATE           limit socket I/O bandwidth
        --bwlimit=RATE
               "--bwlimit=1.5m").  If no suffix is specified, the value 
will be
        --bwlimit=RATE
               specify  a  smaller --bwlimit value, but no larger value 
will be


But, that will not fix SATA hardware problems.


> Or do I have a pile of duff SSD's?


I doubt it.


David

Back to linux.debian.user | Previous | Next — Previous in thread | Find similar | Unroll thread

Thread

new thread, are gigastone ssd's good/bad gene heskett <gheskett@shentel.net> - 2024-01-14 22:50 +0100
  Re: new thread, are gigastone ssd's good/bad David Christensen <dpchrist@holgerdanske.com> - 2024-01-15 09:40 +0100

csiph-web